W3C home > Mailing lists > Public > public-esw-thes@w3.org > February 2010

Re: Doubts about domains and microthesauri in Thesaurus to SKOS conversion

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Sun, 21 Feb 2010 12:12:45 +0100
Message-ID: <4B81152D.80602@few.vu.nl>
To: Johan De Smedt <Johan.De-smedt@tenforce.com>
CC: "L.Will@willpowerinfo.co.uk" <L.Will@willpowerinfo.co.uk>, Giacomo Bartoloni <geco82@gmail.com>, "public-esw-thes@w3.org" <public-esw-thes@w3.org>, Enrico Francesconi <francesconi@ittig.cnr.it>

By the way hasn't this problem be discussed already on the list, for instance at [1], [2] or [3]?
It would be nice to know whether the patterns there can fit in the discussion you have now. I think they could.

By the way like Johan I don't like alternative 1.
And in the STITCH project (for the RAMEAU vocabulary at [4]) we have implemented something that relates a bit to alternative 2. Both "super-vocabulary"and "sub-vocabulary" are instances of skos:ConceptScheme, but it is to the "super-vocabulary" that we have attached some extra information, not to the "sub-vocabulary". We use an extra class ConceptSchemeGrouping for that, which is a sub-class of skos:ConceptScheme and that denotes the notion of a "family" of concept schemes.

Anyway, coming back to Giacomo, original question, there's no way to represent micro-thesauri 'beautifully' in standard SKOS!



[1]http://lists.w3.org/Archives/Public/public-esw-thes/2008Jan/0097.html, continued at http://lists.w3.org/Archives/Public/public-esw-thes/2008Feb/0002.html
[2] http://lists.w3.org/Archives/Public/public-esw-thes/2009Jan/0036.html, continued at http://lists.w3.org/Archives/Public/public-esw-thes/2009Feb/0001.html
[3] http://lists.w3.org/Archives/Public/public-esw-thes/2009Jul/0000.html
[4] http://stitch.cs.vu.nl/rameau

> Three alternatives and some considerations:
> 1)
> The modeling of sub-thesauri in skos would be facilitated if a skos:Collection need not be disjoint with a skos:ConceptScheme.
> This would allow for different sub-classes of skos:Collections
> - collections that represent arrays
> - collections that represent sub concept schemes or ISO like concept groups.
> Note: I would NOT favor this because
> - the skos:member relation mixes the skos:Collection and skos:Concept members of a collection.
> - skos:Collection and skos:ConceptScheme scheme are disjoint classes
>    Thus making a micro-thesaurus a skos:Collection would effectively not allow to model a micro-thesaurus as a skos:ConceptScheme.
>    Occasionally this is not helpful because a micro-thesaurus does is a thesaurus when taken on its own.
> - it would mean a change of the recommendation (while other alternatives are available - see 2)
> 2)
> A new class ConceptGroup may be specified:
> - Having subGroup/superGroup hierarchy possibilities next to membership
> - A Concept Group would not be disjoint from skos:ConceptScheme
> Note: this would be more close to the ISO DIS model Leonard refers to.
> Considerations:
> If some skos extension would define skos-ext:ConceptGroup, with:
> - skos-ext:subGroup/superGroup properties establishing the hierarchy between these groups
> - a skos-ext:inConceptGroup establishing concept group membership,
> then the EUROVOC ontology can take that into account as follows:
> - ev:Domain and ev:MicroThesaurus would be sub-class of skos:ConceptGroup
> - the properties ev:domain/ev:microThesaurus structuring the Domain/Microthesaurus hierarchy would become sub properties of skos-ext:subGroup/superGroup
> -<a-skos-ext:ConceptGroup-uri>  skos:inScheme<a-skos:ConceptScheme-uri>  declares the subject skos-ext:ConceptGroup to be defined by the object skos:ConceptScheme
> 3)
> A separate top-level taxonomy/concept-scheme can be defined to establish the domain/micro-thesaurus (or any thesaurus top-level organization) hierarchy.
> A sub-property of dc:subject (e.g. skos-ext:subjectMatch) could be applied to map the concepts of the original thesaurus to concepts of top-level (domain/micro-thesaurus) thesaurus.
> The sub-property would need to define the semantics of concept-group membership.
> This approach
> - allows multiple classification hierarchies to be applied to an existing thesaurus.
> - makes the original thesaurus independent from the used top-level classification hierarchy.
> - looses the semantics that a top-level concept (domain or microthesaurus) actually can be considered as a thesaurus.
>    (at least in the skos universe because skos:Concept and skos:ConceptScheme are disjoint)
> I would not favor this approach for a thesaurus top level organization.
> - This alternative belongs more on the problem area of matching and mapping thesauri.
> - skos core has a nice arsenal of (matching) properties to handle this problem area.
> - The top-level organization of thesauri can be handled as explained in 2.
> Johan.
>> -----Original Message-----
>> From: public-esw-thes-request@w3.org [mailto:public-esw-thes-
>> request@w3.org] On Behalf Of Leonard Will
>> Sent: 20 February, 2010 19:04
>> To: Giacomo Bartoloni
>> Cc: public-esw-thes@w3.org; Enrico Francesconi
>> Subject: Re: Doubts about domains and microthesauri in Thesaurus to
>> SKOS conversion
>> On 2010-02-18 15:11, Giacomo Bartoloni wrote:
>>> Hi everybody, I'm studying the best way (or the most beautiful and
>>> meaningful form) to represent a thesaurus with domains and
>>> microthesauri, in my case Eurovoc and few ones, in "SKOS 2009"
>> format.
>>> My doubt is how to deal with domains and microthesauri, I found many
>>> examples (UKAT's, IVOA vocabularies and STW's) but it seems to lack a
>>> shared method to do it.
>>> In UKAT thesaurus both the thesaurus and the microthesauri are
>>> skos:ConceptScheme and the concepts have the property skos:inScheme
>>> pointing to the main scheme and to the related microthesaurus.
>>> According to IVOA all the domains and the microthesauri are
>>> skos:Concept but I think that in this way I loose the base thesaurus
>>> Descriptor goal and I put all domains, microthesauri and descriptors
>>> on the same level.
>>> In the STW's thesaurus, there is only one skos:ConceptScheme, the
>>> thesaurus itself, and the domains, the microthesauri and the
>>> descriptors are mapped into skos:Concept but this class is extended
>>> into stw:Descriptor and ztw:Thsys.
>>> Which is the best thesaurus representation?
>>> Thanks a lot
>>> Giacomo Bartoloni
>> You might be interested to see the draft ISO 25963-1 data model, which
>> I
>> attach. This contains some elements that are not (yet) implemented in
>> SKOS, such as ThesaurusArray and ConceptGroup.
>> Arrays are sets of sibling terms, often grouped or ordered by some
>> "characteristic of division" specified in a "NodeLable", whereas
>> ConceptGroups are collections of concepts which may not be
>> hierarchically related to each other, but may be subsets of the whole
>> thesaurus, such as microthesauri or domains.
>> These are described in more detail in the draft standard, which is
>> available for comment until 28th February 2010 on the British Standards
>> Institution web site at<http://drafts.bsigroup.com/>  (Free
>> registration
>> required). Have a look at paragraphs 15.2.17 and 15.2.18.
>> I hope that someone will come up with a way of incorporating these and
>> other parts of the ISO model into SKOS as extensions or additions if
>> necessary. I haven't yet worked out whether the The EUROVOC Thesaurus
>> Ontology Schema that Johan de Smedt sent is consistent with the ISO
>> model, but would be glad so see any comments.
>> Leonard Will
>> --
>> Willpower Information     (Partners: Dr Leonard D Will, Sheena E Will)
>> Information Management Consultants            Tel: +44 (0)20 8372 0092
>> 27 Calshot Way                              L.Will@Willpowerinfo.co.uk
>> ENFIELD                                Sheena.Will@Willpowerinfo.co.uk
>> EN2 7BQ, UK                            http://www.willpowerinfo.co.uk/
Received on Sunday, 21 February 2010 11:13:22 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:46:00 UTC